deferred / deferrable task are not take in account by max_active_tasks

### Apache Airflow version

2.9.2

### What happened?

If you set max_active_tasks, it do not limit the number of concurrent deferred tasks.

It's possible to restrict the number of concurrent deferred tasks with a pool but I want limit the number of tasks run by a dag and keep the possibility for every task to use a specific pool ( we can only use 1 pool by task in airflow )

### What you think should happen instead?

max_active_tasks should take in account the deferred task or we should create a new boolean setting

`max_active_tasks_include_deferred` like what was done for the pools

### How to reproduce

```python
from datetime import datetime, timezone, timedelta

from airflow.sensors.time_sensor import TimeSensorAsync
from pendulum import today
from airflow import DAG

with DAG(
dag_id='example',
schedule_interval='0 0 * * *',
max_active_tasks=10,
start_date=today("UTC").add(days=-1)):
for i in range(100):
TimeSensorAsync(task_id=f"a_{i}", target_time=(datetime.now(tz=timezone.utc) + timedelta(minutes=10)).time())

```

### Are you willing to submit PR?

- [ ] Yes I am willing to submit a PR!

### Code of Conduct

- [X] I agree to follow this project's [Code of Conduct](https://github.com/apache/airflow/blob/main/CODE_OF_CONDUCT.md)

Start by running the provided DAG with max_active_tasks=10 and TimeSensorAsync tasks, then inspect how the scheduler counts deferred tasks and how pool limits handle them. Done means deferred tasks are included in max_active_tasks, or an agreed max_active_tasks_include_deferred setting provides that behavior, with the concurrency behavior covered by regression tests.

Written by the indexing model from the issue text.